1
This section covers the sources and characteristics of biomedical signals, the performance requirements and design constraints of medical instrumentation, bioelectric potentials and electrodes, physiological transducers, and the biomedical recorders ECG, EEG and EMG.
2
The Generalised Medical Instrumentation System • Every medical instrument follows the same chain: measurand (the physiological quantity) → sensor/transducer → signal conditioning (amplification, filtering) → analogue-to-digital conversion → processing → display, recording, storage and transmission, with calibration, feedback and control around it and a power supply with patient isolation. • Sources of biomedical signals: bioelectric (ECG, EEG, EMG, EOG, ERG — from ionic currents across excitable membranes), bioimpedance (respiration, body composition, cardiac output), biomagnetic (MCG, MEG — extremely weak, requiring SQUID magnetometers), biomechanical (pressure, flow, displacement, force), bioacoustic (heart and breath sounds, Korotkoff sounds, Doppler), biochemical (pO₂, pCO₂, pH, glucose, electrolytes, enzymes) and biooptical (oximetry, plethysmography, fluorescence). • Measurement may be direct or indirect, invasive or non-invasive, continuous or intermittent, and in vivo or in vitro; a further distinction is between sensing, diagnostic, therapeutic and laboratory equipment.
3
Signal Typical amplitude Frequency range ECG 0.5 - 5 mV (surface) 0.05 - 100 Hz (diagnostic);
4
0.5-40 Hz monitoring EEG 10 - 100 μV 0.5 - 100 Hz (delta, theta, alpha, beta, gamma) EMG 0.1 - 5 mV 10 - 2,000 Hz EOG 10 - 3,500 μV DC - 100 Hz ERG 0.5 μV - 1 mV 0.2 - 200 Hz Blood pressure 0 - 400 mmHg DC - 50 Hz Respiratory rate 2 - 50 breaths/min 0.1 - 10 Hz Body temperature 32 - 40 °C DC - 0.1 Hz Performance Requirements • Static characteristics: accuracy (closeness to the true value — a measure of systematic error); precision/repeatability (closeness of repeated readings to one another — a measure of random error); resolution (the smallest detectable change); sensitivity (output per unit input, the slope of the calibration curve); linearity; hysteresis; drift (zero and sensitivity drift with time and temperature); range and span; threshold and dead zone; and reproducibility.
5
An instrument may be precise but inaccurate, and precision is the prerequisite for accuracy but does not guarantee it. • Dynamic characteristics: frequency response and bandwidth, rise time, settling time, time constant, damping ratio (an underdamped system overshoots and rings; an overdamped one is sluggish;
6
0.6-0.7 is the usual optimum), natural frequency and phase shift. • Statistics of measurement: a set of readings is described by its mean, median and mode, and its scatter by standard deviation and variance, with the standard error of the mean = σ/√n.
7
Errors are systematic (bias, correctable by calibration) or random (scatter, reduced by averaging n readings, since random error falls as 1/√n), with gross (human) errors a third category.
8
Diagnostic performance uses sensitivity (true positive rate), specificity (true negative rate), positive and negative predictive value and the ROC curve. • Noise and interference:
9
50 Hz mains interference (the dominant problem, reduced by high CMRR, a driven right-leg circuit, shielded twisted leads, good electrode contact and a notch filter), motion and baseline wander, EMG contamination, electrode half-cell drift, thermal and shot noise, and electrosurgical interference.
10
The figure of merit is the signal-to-noise ratio, improved by filtering, averaging and differential recording.
11
Intelligent Systems and Design Constraints • Intelligent (smart) medical instrumentation incorporates a microprocessor or embedded system, giving automatic calibration and self-test, digital filtering and artefact rejection, automatic measurement and computation of derived parameters, alarms with limits, trending and storage, networking and telemetry, user interfaces and fail-safe operation, and increasingly machine-learning-based interpretation and decision support — with automated ECG interpretation the oldest example. • General constraints in the design of medical instrumentation — the standard examination list:
12
(1) the measurand is usually inaccessible and measurement is often indirect;
13
(2) signals are very small and buried in noise and interference;
14
(3) biological variability between and within patients, so 'normal' is a range, not a value;
15
(4) the measurement must not disturb the variable being measured (loading effect);
16
(5) patient safety is paramount — electrical, thermal, mechanical, radiation, infection and chemical hazards;
17
(6) the environment is hostile — motion, fluids, temperature and humidity, electromagnetic interference, and the need for sterilisation;
18
(7) reliability, since failure may be fatal, requiring redundancy and alarms;
19
(8) ease of use by clinical, not engineering, staff;
20
(9) regulatory and standards compliance; and (10) cost, maintenance and calibration. • Electrical safety is the most examinable of these.
21
Macroshock is current through the intact skin:
22
1 mA perception, 10-20 mA 'let-go' (sustained muscle contraction), 100 mA ventricular fibrillation.
23
Microshock is current delivered directly to the heart through an intracardiac catheter or pacing lead, where as little as 10 μA can cause fibrillation, which is why the CF (cardiac floating) classification exists alongside BF (body floating) and B.
24
Protection rests on protective earthing, double insulation, equipotential bonding, isolated (floating) patient circuits with optical or transformer isolation, isolated power supplies with line isolation monitors, limits on leakage current, and regular electrical safety testing under IEC 60601.
25
Bioelectric Potentials • As set out in Chapter 1: the resting membrane potential of about −70 to −90 mV arises from the Na⁺/K⁺ pump, the selective permeability of the membrane and impermeant intracellular anions, and is described by the Nernst equation for a single ion and the Goldman equation for the whole membrane.
26
An adequate stimulus depolarises the membrane to threshold (about −55 mV), whereupon voltage-gated Na⁺ channels open (depolarisation to about +30 mV), then inactivate as K⁺ channels open (repolarisation), often with a brief hyperpolarisation. • The action potential is all-or-none, information being coded in frequency, and is followed by absolute and relative refractory periods.
27
It propagates by local circuit currents — continuously in unmyelinated fibres and by saltatory conduction between nodes of Ranvier in myelinated fibres — the velocity rising with fibre diameter and myelination. • What an electrode on the skin records is not the action potential itself but the volume-conducted field produced by many cells acting together, attenuated and low-pass filtered by the intervening tissue — which is why surface signals are in the millivolt and microvolt range and why electrode placement matters so much.
28
Electrodes • An electrode is a transducer that converts the ionic current of the body into the electronic current of the instrument, by means of the chemical reactions occurring at the electrode-electrolyte interface. • Half-cell potential is the DC potential developed at that interface; a difference between two electrodes produces an offset that can saturate a high-gain amplifier, and drift in it produces baseline wander.
29
Electrodes are polarisable (behaving like a capacitor, no net charge crossing — platinum; large half-cell potential and motion artefact) or non-polarisable (behaving like a resistor, charge crossing freely — silver/silver chloride).
30
Ag/AgCl is therefore the standard biopotential electrode, with a stable, low half-cell potential and low motion artefact.
31
The equivalent circuit of an electrode is a half-cell voltage source with a parallel RC (double layer) in series with the electrolyte/skin resistance. • Skin preparation matters more than the electrode: the stratum corneum is the principal source of impedance and of motion artefact, so the skin is cleaned, lightly abraded and covered with electrolyte gel; impedance should be below about 5 kΩ for ECG and below 5-10 kΩ for EEG, and differences in impedance between electrodes are worse than high impedance itself because they degrade CMRR.
32
Electrode type Description and use Plate/limb electrodes Metal plates strapped to the limbs — the original ECG electrodes Suction (bulb) electrodes For the chest leads of a resting ECG; quick to apply and move Floating (recessed) electrodes Ag/AgCl pellet recessed in a gel-filled cavity so that the metal never touches the skin — the standard disposable ECG and monitoring electrode, with minimal motion artefact Disposable pre-gelled adhesive electrodes Foam or hydrogel, single use — prevents cross-infection Dry and capacitive electrodes No gel; used in wearables and long-term monitoring, with higher impedance EEG scalp electrodes Small cup (disc) electrodes with collodion or paste, placed by the international 10-20 system; also needle and subdural strip/grid electrodes Needle electrodes Inserted through the skin for EMG (concentric or monopolar) and intraoperative monitoring Microelectrodes Tip diameter 0.5-5 μm, to record from inside a single cell; either metal (tungsten or platinum-iridium, etched and insulated except at the tip) or glass micropipettes filled with 3 M KCl.
33
They have very high impedance (megohms to hundreds of megohms), so they demand an amplifier of extremely high input impedance, usually a unity-gain FET buffer close to the tip; they are also noisy and fragile Internal and catheter-tip electrodes Intracardiac, oesophageal, pacing and defibrillation electrodes Stimulating electrodes Deliver current rather than record — pacing, defibrillation, TENS, functional electrical stimulation Physiological Transducers • A transducer converts one form of energy into another — for measurement, usually a physical variable into an electrical signal.
34
Active (self-generating) transducers produce their own output and need no external excitation — piezoelectric, thermocouple, photovoltaic and electromagnetic/induction types.
35
Passive (modulating) transducers require an external excitation supply and vary a passive parameter — resistive (strain gauge, potentiometer, thermistor, RTD, photoresistor), capacitive and inductive (LVDT) types. • Displacement, position and motion: the resistive potentiometer (simple, robust, large output, but with friction and wear); the strain gauge, whose gauge factor = (ΔR/R)/(ΔL/L) is about 2 for metal foil and 50-200 for semiconductor types, used in a Wheatstone bridge with temperature compensation; the LVDT (linear variable differential transformer) — a primary with two opposed secondaries and a movable core, giving frictionless, high-resolution, essentially infinite-life displacement measurement with a phase-sensitive detector to indicate direction; capacitive transducers (C = εA/d); and for motion, accelerometers (piezoelectric or MEMS), tachometers and optical/magnetic encoders. • Pressure transducers: the sensing element is a diaphragm, bellows, Bourdon tube or capsule whose deflection is measured by a strain gauge, LVDT, capacitive or piezoresistive element.
36
Direct (invasive) measurement uses a fluid-filled catheter coupled to an external transducer — cheap and re-zeroable but with damping, resonance and air-bubble problems, and the transducer must be levelled at the phlebostatic axis — or a catheter-tip (micromanometer) transducer, with a far better frequency response but at higher cost and fragility.
37
Indirect measurement is by sphygmomanometry (5.2).
38
Important adjuncts are zeroing to atmosphere, flushing systems and the square-wave damping test. • Photoelectric transducers: photovoltaic cells (generate a voltage — active), photoconductive cells/LDRs (resistance falls with light — passive), photodiodes (fast, linear, used in photovoltaic or reverse-biased photoconductive mode), phototransistors (more sensitive, slower) and photomultiplier tubes (extremely sensitive to very low light, used in gamma cameras, flow cytometers and luminometers).
39
Their medical applications are the basis of pulse oximetry, photoplethysmography, colorimetry and spectrophotometry, blood cell counters, bilirubinometers and optical fibre sensors. • Other transducers: thermistor (a semiconductor with a large negative temperature coefficient — very sensitive but non-linear, the usual clinical temperature sensor), RTD/platinum resistance thermometer (linear and accurate, used as a standard), thermocouple (Seebeck effect, active, small and fast, needing a reference junction), piezoelectric (ultrasound transducers, phonocardiography, force), electromagnetic flow transducers and chemical/biosensors.
41
ECG, EEG and EMG • Common front end: all three require an instrumentation amplifier built from three op-amps, with very high input impedance (> 10 MΩ, and far higher for microelectrodes), high differential gain, low noise and drift, and a very high common-mode rejection ratio (> 100 dB, i.e. > 10⁵:1) to reject the mains interference that is common to both inputs.
42
This is followed by isolation (optical or transformer) for patient safety, a defibrillator protection circuit, filters, an ADC and the display or recorder.
43
The driven right-leg circuit feeds the inverted common-mode signal back to the patient, further improving rejection. • Electrocardiograph (ECG): records the electrical activity of the heart (1.4).
44
12 leads — three bipolar limb leads I, II and III (Einthoven's triangle, where lead II = lead I + lead III), three augmented unipolar limb leads aVR, aVL and aVF derived against Wilson's central terminal, and six unipolar chest leads V1-V6.
45
Standard settings are a sensitivity of 10 mm/mV and a paper speed of 25 mm/s, so one small (1 mm) square is 0.04 s and one large (5 mm) square 0.2 s.
46
Waves and normal intervals:
47
P (atrial depolarisation), QRS (ventricular depolarisation, < 0.12 s), T (ventricular repolarisation), PR interval 0.12-0.20 s, QT interval.
48
Uses: arrhythmia, ischaemia and infarction, chamber hypertrophy, electrolyte disturbance and drug effects.
49
Artefacts: mains interference, baseline wander from respiration and motion, muscle tremor, loose electrodes and lead reversal. • Electroencephalograph (EEG): records cortical activity from scalp electrodes placed by the international 10-20 system, in which electrode spacing is 10 % or 20 % of the distance between the landmarks nasion, inion and the preauricular points, with odd numbers on the left, even on the right and z for the midline.
50
Signals are only 10-100 μV, so amplification of about 10⁶ and meticulous technique are needed; recording may be bipolar or referential (monopolar), in montages.
51
Rhythms: delta < 4 Hz (deep sleep, pathological when awake), theta 4-8 Hz (drowsiness, children), alpha 8-13 Hz (relaxed wakefulness with eyes closed, maximal occipitally and abolished by eye opening), beta 13-30 Hz (alert, mental activity, drug effect) and gamma > 30 Hz.
52
Uses: epilepsy (the principal indication), encephalopathy, sleep studies, coma and brain-death assessment, depth of anaesthesia and evoked potentials.
53
Artefacts: eye blink and movement, ECG, EMG from scalp muscles, sweat, electrode pop and mains. • Electromyograph (EMG): records the electrical activity of skeletal muscle, using surface electrodes (non-invasive, for kinesiology, biofeedback, prosthesis control and gait analysis, but with cross-talk and poor selectivity) or needle electrodes (concentric or monopolar, for clinical diagnosis, recording individual motor unit action potentials).
54
Signals are 0.1-5 mV over 10-2,000 Hz, are inherently stochastic and biphasic, and are assessed at rest (where spontaneous fibrillation potentials indicate denervation), on minimal contraction (motor unit morphology — small short polyphasic units in myopathy, large long units in neuropathy) and on maximal effort (recruitment and interference pattern).
55
EMG is normally accompanied by nerve conduction studies, which measure conduction velocity = distance ÷ (proximal latency − distal latency), normally about 50-60 m/s in large motor nerves, reduced in demyelination while amplitude falls in axonal loss.